Ignore non-standard DAP events to ensure debugging can start

[pieterdd]

As per #11777, this issue caused Helix to stop debugging when launching a debugpy DAP session:

2024-09-25T15:40:33.914 helix_dap::transport [ERROR] err: <- Parse(Error("unknown variant debugpySockets, expected one of initialized, stopped, continued, exited, terminated, thread, output, breakpoint, module, loadedSource, process, capabilities, memory", line: 0, column: 0))

This fix allows a debugpy DAP session to start up. Although there are other issues preventing debugpy to be used in conjunction with Helix, this is a step forward.

[the-mikedavis]

I don't think we should be discarding all parsing failures. If the Event type is flexible in what it accepts like Request/Response it should be written as a trait instead of an enum. It's very similar in the spec to LSP's notification type which is also written as a trait (helix-lsp-types/src/notification.rs).

[the-mikedavis]

I've pushed up fec5101 which moves the helix_dap code closer to what we have in helix_lsp, including logging and discarding unknown events. I tried the reproduction case from #11777 and see the same behavior between these changes and mine - discarding the notifications gets further in the DAP session but it still crashes out - so I will close this out.